Author |
Message |
Vilas Vyankatrao Jangale (vilasmechgmailcom)
Intermediate Member Username: vilasmechgmailcom
Post Number: 20 Registered: 5-2010
| Posted on Monday, August 30, 2010 - 5:53 pm: | |
Hi All, Guruditya has clearly mentioned in his post that the equation is to compute spectral irradiance. Guruditya - Can you please specify the wavelength range that this spectrometer works in? As far as know, in NIR, water has a very strong absorption and carbon dioxide has relatively very weak absorption. Also, the absorption peak of carbon dioxide at 1400 nm is completely overlapped by water. What chemometric methods are you using to characterize the mixtures? I look forward to your reply. Thank you, Vilas |
Howard Mark (hlmark)
Senior Member Username: hlmark
Post Number: 342 Registered: 9-2001
| Posted on Monday, August 30, 2010 - 11:07 am: | |
Daniel - makes a good point. Those of us who frequent this discussion board automatically think of "calibration constants" as meaning the coefficients needed to turn measured spectral data into analytical results. But it doesn't necessarily mean that to everybody, especially to the electrical engineers who typically design the instrumentation. The nature of the "calibration constants" that Guruditya described should have made us suspicious, at least. Guruditya's first task, therefore, is to contact the manufacturer and find out exactly what information he's missing, since our automatic assumption about that might be incorrect. If that turns out to be the situation, then he REALLY has a case to make for fraud in advertising, I think. \o/ /_\ |
Daniel Alomar (dalomar)
New member Username: dalomar
Post Number: 5 Registered: 2-2009
| Posted on Monday, August 30, 2010 - 9:41 am: | |
Hello all, It seems to me that the manufacturer is informing the calibration method for the instrument irradiation and not the method to obtain an equation to predict CO2 or whatever... If the objective is (and if this is not the case, I apologize) to develop a calibration for prediction purposes, a proper diagnostics routine and decent chemometric software are more important than obtaining the instrument calibration constants. Regards, Daniel |
David W. Hopkins (dhopkins)
Senior Member Username: dhopkins
Post Number: 160 Registered: 10-2002
| Posted on Saturday, August 28, 2010 - 9:12 pm: | |
Hi Venky, Are you be a neighbor to Guruditya? Perhaps you can help with some more standard methods of calibration? I don't understand, does the user have to calibrate the instrument? Or, is the instrument supplied with the calibrations for CO2 and H2O, and the question was just for information? I hope it is the latter.... Best regards, Dave |
venkatarman (venkynir)
Senior Member Username: venkynir
Post Number: 114 Registered: 3-2004
| Posted on Saturday, August 28, 2010 - 8:49 pm: | |
Dave & Mark I gone through papers , nothing is mentioned in it. In INDIA there is no direct sales it might be either dealers or through others only. I also suggested to combination of experimental and theoretical calculation . |
Howard Mark (hlmark)
Senior Member Username: hlmark
Post Number: 341 Registered: 9-2001
| Posted on Saturday, August 28, 2010 - 3:15 pm: | |
Guruditya - first of all, if you haven't actually bought the instrument yet, or haven't completed paying for it, you have some financial and legal leverage to force the manufacturer to tell you what the constants are, that you need. I don't know what the laws are in India, but in the USA the manufacturer would be guilty of false advertising, to tell you that you can measure H2O and CO2, when you really can't because you don't have the critical information. In any case, you could withold payment until you get the info you need, and apply financial pressure that way. As for the technical aspects of developign your own constants, what David is telling you is correct, but may be incomplete, because you may not be telling us some critical facts. Another thing we (and you) need to know is whether anything else is in the mixtures that constitute the samples, and whether they might also vary. If that is the case, then you'll probably need more than eight samples, even using the MLR algorithm for the calibrations. Howard \o/ /_\ |
David W. Hopkins (dhopkins)
Senior Member Username: dhopkins
Post Number: 159 Registered: 10-2002
| Posted on Saturday, August 28, 2010 - 2:28 pm: | |
Hi Guruditya, One other observation, with 8 calibration constants, you would have to use at least 8 different samples, with various concentrations and proportions of the CO2 and H2O. Best wishes, Dave |
David W. Hopkins (dhopkins)
Senior Member Username: dhopkins
Post Number: 158 Registered: 10-2002
| Posted on Saturday, August 28, 2010 - 2:24 pm: | |
Hi Guruditya, This is certainly a unique method for calibrating a spectrometer, no matter what kind, NIR or other. It looks to me as if the manufacturer wants to divulge as little as possible about the calibration method. I expect that each of the additive terms would concern a different wavelength. Do you know how many wavelengths are measured in this spectrometer? Is it a filter or monochromator instrument? I suppose that you could just use MLR to determine the coefficients, once you have the wavelengths selected. It might be messy, though. I'll be interested to hear the impression of others. Best regards, Dave |
Guruditya Sinha (guruditya)
New member Username: guruditya
Post Number: 1 Registered: 8-2010
| Posted on Saturday, August 28, 2010 - 6:04 am: | |
I have an NIR spectrometer for the detection of CO2 and water vapour. The manufacturers have published a paper on this, and in it they have indicated that calibrations have been performed, with spectral irradiance calculated as follows: Y=1/(w^5)exp(a+b/(w))*(c+(d/w)+(e/(w^2))+(f/(w^3))+(g/(w^4))+(h/(w^5))) where w = wavelength and a,b,c,d,e,f,g,h, are the calibration constants. These have not been provided by the manufacturer. The query arises as to how one can calculate effectively the calibration constants. Along with some peers I have tried to calculate this on the basis of some test results( two in number for only two wavelengths) provided by the manufacturer. But this may not be appropriate as these would obviously not be enough for calculating all the constants. Can some theoretical approach be applied for somehow obtaining a desired result, i.e., the calibration constants? |
|